Cortensor

🔎 Recap: How the Cortensor Stack Has Been Built Layer by Layer One thing that may not always be obvious from individual DevLogs is that we are not trying to build every part of the stack at the same depth at the same time. The approach has been much more sequential. 🔹 Infrastructure first - network / node infrastructure - validators, routers, and sessions - dashboard / indexer - Testnet → Mainnet Lite → Mainnet Full / L3 - recovery and operational tooling These paths reached a working MVP / E2E baseline first and continue to be hardened as usage grows. 🔹 Then the product layer - Portal became the simpler hosted API entry point into the network - Corgent / Bardiel added delegation, validation, fact-checking, and trust-oriented product surfaces - These products continue improving while the infrastructure underneath is expanded and refined 🔹 Current product spearhead A lot of the active iteration is now around PyClaw + the Desktop App. - PyClaw has reached its first 0.1 baseline - Research, coding, memory, tools, scheduling, delegation, recovery, verification, and longer-running workflows are now part of the core runtime at different levels of maturity - The Desktop App is being developed as another interface and broader distribution layer for that same runtime 🔹 The broader progression Core infrastructure → Mainnet + product integration → Portal / Corgent / Bardiel access → PyClaw agent runtime → Desktop + broader distribution → real developer and user workloads → continuous improvement 🔹 Why “100% complete” is difficult to define - Mainnet Full can be working while still expanding capacity and operational reliability - Portal can be functional while still improving models, compatibility, onboarding, and scale - PyClaw can reach 0.1 while continuing to improve quality, autonomy, workflows, and Desktop distribution These are maturity milestones rather than permanent finish lines. 🔹 Current takeaway - The strategy is not to spread effort across unrelated products at once - It is to establish one layer, get it working E2E, connect the next layer, and then keep improving each part as real usage exposes new gaps - Mainnet Full, Portal, Corgent, Bardiel, PyClaw, and Desktop are increasingly becoming connected layers of the same stack #Cortensor #MainnetFull #Portal #PyClaw #Corgent #Bardiel #AIInfra

🛠️ DevLog – PyClaw: 0.1.15 Is Taking Shape Across Security, Scheduling, Memory, and MCP The next PyClaw patch release is starting to come together, with a broader set of runtime, security, scheduling, storage, and interoperability improvements already confirmed. Expected version: 0.1.15 🔹 Security workflow expansion - Local security audits are becoming more isolated and repeatable - Recurring evidence collection has been expanded - SARIF export is now part of the workflow - Offline AWS checks and bounded web / API scanning are being added with tighter execution limits 🔹 Tasks, cron, and scheduled execution - Task and cron controls now cover more of the full lifecycle - Run-now, scheduler status, history, one-shot execution, and watch delivery are being tightened - This gives scheduled workflows better visibility and control instead of treating them as fire-and-forget jobs 🔹 Artifact and retrieval durability - Artifact retention validation has been strengthened - Alias expiry handling is more reliable - Retrieval ingestion is becoming more durable across longer-lived workflows 🔹 PostgreSQL and storage reliability - PostgreSQL-backed APIRuntime startup is now supported more cleanly - Storage migration failures provide clearer output - This continues reducing ambiguity around startup, migration, and recovery behavior 🔹 Memory and recovery - Memory consolidation now receives more trusted runtime context - Capture recovery is becoming more capability-aware - The goal is to make memory writes and recovery paths more explicit and dependable 🔹 Telemetry and debugging - Exporter behavior has been tightened - Debugging visibility has improved - Token history is linked more clearly into runtime telemetry 🔹 MCP interoperability - HTTP / SSE compatibility has been strengthened - Coverage has been updated against the current official SDK behavior - This continues improving PyClaw's ability to work with a broader set of external MCP servers and tools 🔹 Still in flight - Room activation locking - Failed-runtime-start cleanup - Memory-write authorization - These changes are green individually but are not yet part of the confirmed release set 🔹 Release status - The branch is not release-ready yet - The latest integrated CI has one remaining macOS Focus PTY resize failure - The rest of the hosted checks passed - That final failure still needs to be rerun or fixed before 0.1.15 can be promoted 🔹 Current takeaway - 0.1.15 is shaping up as another broad reliability release across security workflows, scheduled execution, memory, storage, artifacts, telemetry, and MCP - The release will move forward once the remaining integration issue is cleared and the final set is locked #Cortensor #DevLog #PyClaw #MCP #Cybersecurity #AgenticAI #DeveloperTools

PyClaw + Portal + ongoing hackathons are starting to form a practical builder loop. 🔹 PyClaw gives developers the agent runtime - tools - memory - delegation - streaming - structured output - longer-running workflows 🔹 Portal gives builders the hosted model path - simpler API access - Gemma4 + other models as capacity allows - real testing around compatibility, latency, reliability, and model behavior 🔹 Hackathons connect both - bring in more external builders - create real agent workloads - expose SDK, Portal, model, onboarding, and runtime gaps - turn those findings directly into product fixes The feedback loop is simple: PyClaw → real workflows → Portal / models → feedback → product fixes → stronger builder experience → more developers The goal is not just more submissions. It is continuous real-world iteration across PyClaw, Portal, models, and the broader Cortensor stack. #Cortensor #PyClaw #Portal #Gemma4 #Hackathon #SDK #AgenticAI

🛠️ DevLog – PyClaw: 0.1.14 Improves Time Grounding, Typed Tools, Storage Safety, and Recovery The latest PyClaw release is now out, continuing the 0.1.x focus on more reliable agent execution, stronger tool contracts, and safer state / storage management. 🔹 More reliable time grounding - Exact UTC dates can now be grounded through canonical time evidence - This gives workflows a stronger source of truth when current date context matters - It also reduces ambiguity across longer-running and resumed tasks 🔹 Stronger typed tools - Tool handlers now support typed Pydantic and dataclass inputs - Structured arguments can move through tool execution with clearer schemas and validation - This makes more complex SDK and agent-tool integrations easier to define consistently 🔹 Memory and resource cleanup - Memory resource ownership and cleanup have been tightened further - PostgreSQL-backed workflows receive stronger cleanup behavior - This continues reducing resource leakage across persistent and longer-running runtimes 🔹 Safer storage operations - Storage commands now provide clearer exit-state behavior - Backup and rollback paths have been strengthened - Failure states are easier to distinguish before potentially destructive storage operations continue 🔹 Security and recovery - Security evidence validation has been tightened - Plan recovery is more reliable when workflows are interrupted or need to resume from existing state - The focus remains evidence-backed execution rather than assuming a step succeeded 🔹 Current takeaway - 0.1.14 continues tightening the runtime underneath longer and more structured agent workflows - Time grounding, typed tools, storage safety, memory lifecycle, evidence validation, and recovery are all becoming more explicit as the 0.1 line matures #Cortensor #DevLog #PyClaw #SDK #AgenticAI #DeveloperTools #AIInfra

Monthly Node Rewards – September 2026 We've now sent out the September 2026 $COR node rewards to 20 nodes as part of our monthly node reward program. 🔹 Rewards sent to 20 nodes total 🔹 Includes 10 dedicated nodes + 10 ephemeral nodes 🔹 Recognizing uptime, stability, and contribution 💬 Join our Discord to set up a node or get involved: discord.com/invite/gPHh6wtZy… Thanks to all node operators continuing to support the network and push decentralized AI forward - it’s $COR or nothing. #Cortensor #AI #DePIN #NodeRewards #COR

🛠️ DevLog – PyClaw: 0.1.13 Expands Tool Execution, Freshness Checks, Memory Recovery, and Cross-Platform Builds The latest PyClaw release is now out, with another reliability-focused update across bounded tool execution, answer freshness, storage recovery, scheduled memory, and cross-platform packaging. 🔹 Bounded tool execution - Tool execution now supports broader bounded workflows while preserving user-defined limits - This helps agents do more within a task without weakening explicit execution boundaries 🔹 Freshness verification - Direct answers and continuation turns now receive stronger freshness checks - This reduces the chance of continuing from stale state when newer context or results should take priority 🔹 SQLite recovery - SQLite snapshot recovery has been improved across platforms - This strengthens recovery behavior for local runtime state on macOS, Linux, and Windows 🔹 Memory reliability - Scheduled-memory recovery is more robust - Memory administration workflows have been tightened - Audited record cleanup is now more consistent across both SQLite and PostgreSQL 🔹 Distribution updates - Desktop, CLI, and daemon builds have all been refreshed - Updated builds are available across macOS, Linux, and Windows - This continues bringing the runtime and Desktop release paths closer together 🔹 Current takeaway - 0.1.13 continues the 0.1.x focus on making longer-running agent workflows more controlled, recoverable, and portable - Tool limits, freshness, memory lifecycle, storage recovery, and distribution are all being tightened together rather than treated as separate concerns #Cortensor #DevLog #PyClaw #DesktopApp #SDK #AgenticAI #DeveloperTools

🛠️ DevLog – Hackathon #5: Early Participation Reaches 26 Builders A quick progress update as Hackathon #5 continues to gain traction. 🔹 Current participation - 26 builders have joined so far - 7 projects have already been submitted - The submission window remains open through 12/18/2026 🔹 What we are watching - how builders onboard into PyClaw 0.1 - which SDK, tool, memory, delegation, and workflow paths they naturally use - where docs, examples, provider setup, or runtime behavior still create friction - useful bug reports and fix PRs that come out of real project work - Portal / model usage where builders choose to use the hosted path 🔹 Why this matters - The goal is not only the final submission count - We want sustained external usage across the full hackathon window - More builders and more projects give us stronger signals around SDK quality, runtime gaps, onboarding, model behavior, and the kinds of agent applications people actually want to build 🔹 Current takeaway - Early participation has now grown to 26 builders with 7 submissions - We will keep promoting the round, refining the SDK / docs, and triaging external feedback as more projects take shape over the coming weeks DoraHacks: dorahacks.io/hackathon/2418/… #Cortensor #DevLog #PyClaw #Portal #Hackathon #SDK #OpenSource

🛠️ DevLog – PyClaw Desktop: Recent Progress and the Next Release Step A broader recap of recent PyClaw Desktop development as we continue closing the gap between the CLI / runtime and the Desktop experience. These updates are already merged into development, but have not yet been published as tagged Desktop releases. 🔹 Better CLI / Desktop parity - Agent lookup and delegation commands now behave more consistently across surfaces - /copy support has been improved - Skill browsing has expanded, along with additional runtime visibility through /graph status 🔹 Conversation and workflow UX - Conversation search now handles multilingual text input more reliably - Rename dialogs no longer lose newer edits while a save is in progress - Message and code copying now have better focus handling and clearer error feedback - Agent / MCP reports now correctly follow the agent attached to the selected conversation 🔹 Desktop accessibility and navigation - Settings now support full keyboard navigation - Additional interaction and focus improvements continue moving Desktop toward a more complete everyday workflow surface 🔹 Public project readiness - README branding, screenshots, installation guidance, privacy notes, and contribution guidance have been refreshed - Desktop now has its own independent versioning baseline starting at 0.0.1 - The latest development baseline has passed 505 tests across Ubuntu and macOS CI, including native Electron checks 🔹 Moving into tagged Desktop releases - Up to this point, Desktop development has been progressing without public version tags or published releases - Starting with the next release cycle, we plan to begin tagging and publishing Desktop versions alongside the ongoing runtime / CLI release process - This gives Desktop a clearer release history and makes it easier for users and contributors to track changes between builds 🔹 Current takeaway - Desktop is moving from continuous development builds toward a more structured public release process - The goal remains to expose the important PyClaw runtime capabilities through a simpler interface while keeping Desktop and CLI backed by the same underlying runtime #Cortensor #DevLog #PyClaw #DesktopApp #OpenSource #DeveloperTools #AgenticAI

🛠️ DevLog – PyClaw: 0.1.12 Improves Research Recovery, PostgreSQL Memory, and Runtime Reliability The latest PyClaw release is now out, with another reliability-focused pass across research recovery, PostgreSQL-backed memory, deterministic runtime state, tool handling, and resource ownership. 🔹 Resumable research workflows - Research checkpoints now preserve failed synthesis evidence - Interrupted or failed synthesis paths retain more useful state for inspection and continuation - This improves recovery behavior for longer-running research workflows 🔹 PostgreSQL memory and sessions - Memory retention and cleanup have been tightened - Session filtering and lifecycle handling are more reliable - Additional resource cleanup reduces lingering storage connections and ownership ambiguity 🔹 More deterministic runtime state - Event ordering is more predictable - Room participant ordering remains stable when timestamps collide - Fallback tool-call IDs are now more reliably unique across response batches 🔹 Clearer failure handling - Fatal runtime failures are distinguished more clearly from recoverable tool errors - Failed runs are reflected consistently across SDK execution and persisted history - Recoverable tool failures can remain part of the workflow without incorrectly collapsing the full run 🔹 Tool and resource handling - Tool schemas and annotation handling have been tightened - Runtime ownership and cleanup behavior is more explicit - memctl workflows received additional reliability improvements across inspection and storage lifecycle handling 🔹 Upgrade note - Database migrations 14-15 require a backup before upgrading - Coordinated writer shutdown is recommended while applying these migrations 🔹 Current takeaway - 0.1.12 continues tightening the foundations needed for longer-running and recoverable agent workflows - The focus remains predictable state, durable memory, clearer failures, and cleaner runtime ownership as the 0.1 line matures #Cortensor #DevLog #PyClaw #SDK #AgenticAI #DeveloperTools #AIInfra

📌 October Node Rewards Update The October node reward snapshot has now been taken and finalized for 19 eligible nodes. Eligible nodes: - 9 ephemeral nodes - 10 dedicated nodes Rewards are expected to be sent out within the next 48 hours based on this finalized snapshot. We’ll share another update once distribution is complete. Thanks to all node operators continuing to support the network. #Cortensor #NodeRewards #NodeOps #MainnetFull

🗓️ Weekly Focus – Mainnet Full Alpha, Hackathon #5 & PyClaw Iteration This week is largely a continuation of last week: keep Mainnet Full Alpha stable, support Hackathon #5 builders, and use the growing external feedback loop to keep tightening PyClaw 0.1.x. 🔹 Mainnet Full – Alpha Monitoring ‐ Continue controlled Alpha monitoring across Portal, PyClaw, Corgent, Bardiel, routers, validators, quality signals, dashboard, and indexer. ‐ Track reliability, capacity, onboarding, model behavior, and product gaps as more builder workloads come through. 🔹 Hackathon #5 – Builder Support ‐ Continue supporting builders through setup, SDK usage, examples, docs, Portal access, and workflow questions. ‐ Keep using Hackathon #5 as a real external validation layer for PyClaw, Portal, models, and the broader developer experience. dorahacks.io/hackathon/2418/… 🔹 Hackathon #5 – External Feedback Loop ‐ Follow up on bug reports, compatibility findings, feature requests, and PRs coming from external developers. ‐ Keep tightening the builder usage → report → reproduce → fix → regression → patch → public verification loop. 🔹 PyClaw – 0.1.x Bug Fixes & Polish ‐ Continue daily triage and patch releases around regressions, runtime reliability, scheduling, memory, tools, delegation, and Focus behavior. ‐ Prioritize builder-blocking issues and external findings before expanding the core feature surface. 🔹 PyClaw – External PR / Contribution Flow ‐ Review and merge useful external fixes while keeping regression coverage and runtime contracts intact. ‐ Refine the contribution workflow so community PRs can move from report → review → CI → merge → release more predictably. 🔹 PyClaw – Desktop / CLI / SDK Iteration ‐ Continue closing parity gaps across Desktop, CLI, Focus, SDK, models, approvals, schedules, skills, artifacts, and runtime controls. ‐ Keep improving the public builder surface so external developers can reproduce and verify fixes more easily. 🔹 PyClaw – MCP, Scheduling & Memory ‐ Continue hardening MCP lifecycle, scheduled-task controls, memory continuity, and runtime recovery based on recent feedback. ‐ Focus on making these workflows dependable enough for longer-running real projects rather than isolated demos. This week is about keeping Mainnet Full Alpha steady while turning Hackathon #5 and external developer activity into a repeatable product-improvement loop for PyClaw and the broader Cortensor stack. #Cortensor #MainnetFull #Alpha #PyClaw #Portal #Hackathon #AgenticAI #AIInfra #MCP #DePIN

🗓️ Weekly Focus – Mainnet Full Alpha, Hackathon #5 & PyClaw Iteration This week is largely a continuation of last week: keep Mainnet Full Alpha stable, support Hackathon #5 builders, and use the growing external feedback loop to keep tightening PyClaw 0.1.x. 🔹 Mainnet Full – Alpha Monitoring ‐ Continue controlled Alpha monitoring across Portal, PyClaw, Corgent, Bardiel, routers, validators, quality signals, dashboard, and indexer. ‐ Track reliability, capacity, onboarding, model behavior, and product gaps as more builder workloads come through. 🔹 Hackathon #5 – Builder Support ‐ Continue supporting builders through setup, SDK usage, examples, docs, Portal access, and workflow questions. ‐ Keep using Hackathon #5 as a real external validation layer for PyClaw, Portal, models, and the broader developer experience. dorahacks.io/hackathon/2418/… 🔹 Hackathon #5 – External Feedback Loop ‐ Follow up on bug reports, compatibility findings, feature requests, and PRs coming from external developers. ‐ Keep tightening the builder usage → report → reproduce → fix → regression → patch → public verification loop. 🔹 PyClaw – 0.1.x Bug Fixes & Polish ‐ Continue daily triage and patch releases around regressions, runtime reliability, scheduling, memory, tools, delegation, and Focus behavior. ‐ Prioritize builder-blocking issues and external findings before expanding the core feature surface. 🔹 PyClaw – External PR / Contribution Flow ‐ Review and merge useful external fixes while keeping regression coverage and runtime contracts intact. ‐ Refine the contribution workflow so community PRs can move from report → review → CI → merge → release more predictably. 🔹 PyClaw – Desktop / CLI / SDK Iteration ‐ Continue closing parity gaps across Desktop, CLI, Focus, SDK, models, approvals, schedules, skills, artifacts, and runtime controls. ‐ Keep improving the public builder surface so external developers can reproduce and verify fixes more easily. 🔹 PyClaw – MCP, Scheduling & Memory ‐ Continue hardening MCP lifecycle, scheduled-task controls, memory continuity, and runtime recovery based on recent feedback. ‐ Focus on making these workflows dependable enough for longer-running real projects rather than isolated demos. This week is about keeping Mainnet Full Alpha steady while turning Hackathon #5 and external developer activity into a repeatable product-improvement loop for PyClaw and the broader Cortensor stack. #Cortensor #MainnetFull #Alpha #PyClaw #Portal #Hackathon #AgenticAI #AIInfra #MCP #DePIN

🗓️ Weekly Recap – Mainnet Full Alpha, Hackathon #5 & PyClaw Iteration This week kept Mainnet Full Alpha moving while Hackathon #5 brought in more external builders and PyClaw continued a fast 0.1.x polish cycle driven increasingly by real usage and public feedback. 🔹 Mainnet Full – Alpha Progress ‐ Mainnet Full continued moving from infrastructure validation toward controlled product and developer usage. ‐ Builder workloads are now becoming another signal for routing, reliability, model capacity, onboarding, and product gaps ahead of Beta. 🔹 Hackathon #5 – Early Traction ‐ Hackathon #5 reached 26 builders and 7 submitted projects, with submissions open through 12/18. ‐ The round is increasingly functioning as a real external validation layer for PyClaw, Portal, SDK onboarding, docs, models, and developer UX. 🔹 PyClaw – Public Feedback & Bug Triage ‐ Daily issue triage continued as more external developers submitted bugs, fixes, and compatibility findings. ‐ The loop is becoming repeatable: builder usage → report → reproduce → fix → regression → patch release → public verification. 🔹 PyClaw – 0.1.x Reliability ‐ Releases 0.1.7 → 0.1.11 improved long-running projects, delegation/verification, failure evidence, memory continuity, scheduling, event ordering, and runtime cleanup. ‐ Focus remains on hardening the 0.1 baseline rather than rapidly expanding the feature surface. 🔹 PyClaw – Scheduling, MCP & Memory ‐ Scheduled-task controls continued improving across run-now, update, pause/resume, retry, history, delivery, and incident handling. ‐ MCP and memory also matured around OAuth/lifecycle management, live reloads, scoped recall, introspection, recovery health, and longer-horizon continuity. 🔹 PyClaw – Desktop / CLI Parity ‐ Desktop continued closing the parity gap across sessions, tasks, schedules, models, approvals, diagnostics, skills, artifacts, usage, and runtime controls. ‐ The Desktop repository is now public, opening more of the end-user surface to external inspection and contribution. 🔹 PyClaw – Open Distribution Pipeline ‐ CLI/daemon, Desktop, and packaging repositories are now public, exposing more of the source → package → binary → manifest → distribution path. ‐ Additional release cycles exercised verified manifests and unsigned macOS, Linux, and Windows runtime builds. 🔹 Hackathon → Product Feedback Loop ‐ Real builder projects are now helping expose runtime, SDK, Portal, model, docs, and onboarding gaps much earlier. ‐ The broader loop remains: infrastructure → products → builders → applications → usage → feedback → stronger products + capacity. A productive week overall - Mainnet Full Alpha continued broadening into real developer usage, Hackathon #5 gained momentum, and PyClaw’s 0.1 baseline is now being improved through a much more active external feedback loop. #Cortensor #MainnetFull #Alpha #PyClaw #Portal #Hackathon #AgenticAI #AIInfra #MCP #DePIN

📌 October Node Rewards Update October node rewards are expected to be processed within the next 72 hours. We are finalizing the eligible node list and reward calculations now, and will share another update once distribution is complete. Thanks to all node operators continuing to support the network. #Cortensor #NodeRewards #NodeOps #MainnetFull

🛠️ DevLog – PyClaw: Latest Release Improves Event Ordering, Run History, and Runtime Cleanup The latest PyClaw release is now out, with another reliability-focused pass across event ordering, longer run histories, temporary-tool safety, agent handoffs, and runtime resource cleanup. 🔹 Session and history consistency - Event ordering is now deterministic when timestamps collide - History compaction is more stable across longer sessions and storage changes - This helps keep replay, resume, and transcript reconstruction more predictable 🔹 Longer run reconstruction - Run-status reconstruction now scales beyond the earlier 2,000-event boundary - This gives longer-running projects and autonomy workflows a more reliable status / recovery path 🔹 Temporary tool safety - Temporary SDK and handoff tools now carry clearer explicit risk metadata - This makes approval and execution policy easier to reason about when short-lived capabilities are introduced 🔹 Safer agent handoffs - Cross-thread agent handoffs have been tightened - The goal is to reduce state leakage and make concurrent / delegated execution more predictable 🔹 Runtime cleanup - Memory, audit, SQLite, and import-related resources now clean up more reliably - This is especially important for longer-running processes, repeated sessions, and cross-platform runtime stability 🔹 Current takeaway - This release is mostly about making the runtime more deterministic and durable under longer-lived workloads - Event history, status reconstruction, temporary tools, handoffs, and resource lifecycle are all getting tighter as the 0.1 line continues to mature #Cortensor #DevLog #PyClaw #SDK #AgenticAI #DeveloperTools #AIInfra

Monthly Node Rewards – September 2026 We've now sent out the September 2026 $COR node rewards to 20 nodes as part of our monthly node reward program. 🔹 Rewards sent to 20 nodes total 🔹 Includes 10 dedicated nodes + 10 ephemeral nodes 🔹 Recognizing uptime, stability, and contribution 💬 Join our Discord to set up a node or get involved: discord.com/invite/gPHh6wtZy… Thanks to all node operators continuing to support the network and push decentralized AI forward - it’s $COR or nothing. #Cortensor #AI #DePIN #NodeRewards #COR